Futher life of Zevo volumes?

Issues specific to non-OpenZFS pools (i.e., ZEVO & MacZFS).

Futher life of Zevo volumes?

Postby dukzcry » Mon Jul 28, 2014 3:35 am

I had used my Zevo-created volume with O3X since Zevo's decadence. Now i'm deciding whether:
1) Should i create an OpenZFS volume with O3X and migrate my Zevo's data there?
or
2) Should i try to do 'zpool upgrade' on my Zevo's volume? Will it work and is it safe to do? Last time i've tried it with first stable version of O3X, it failed.
or maybe even
3) Stay with my current Zevo-created volume. Will the above two points give me any advantage over this variant?

So what recommendation could O3X authors give me?
Thanks.
User avatar
dukzcry
 
Posts: 2
Joined: Sun Mar 23, 2014 2:38 am

Re: Futher life of Zevo volumes?

Postby ilovezfs » Mon Jul 28, 2014 4:52 am

Any of those options should work.

The following information is almost entirely redundant with what has already been explained in a slightly different context at length here: https://openzfsonosx.org/wiki/Zpool#Creating_a_pool

The possible benefits of rebuilding a pool:

File system properties
1. lz4
[not available in ZEVO and not available in pool version 28; must be specified at the time a dataset is created] Option to enable lz4 compression before putting any data in the pool. The same effect can be achieved in an old pool by creating a new dataset with lz4 enabled, and copying data there. If the data is already compressed or is encrypted, this would yield little benefit.
2. dedup
[not available in ZEVO; must be specified at the time a dataset is created] Option to enable dedup before putting any data in the pool. The same effect can be achieved in an old pool by creating a new dataset with dedup enabled, and copying data there. Dedup requires a lot of memory, and is usually not recommended for home users.
3. ZVOLs
[not available in ZEVO; a type of ZFS dataset, to which a ZFS file-system dataset cannot be converted] Option to use a ZVOL dataset instead of a file-system dataset. A ZVOL presents as unformatted block storage, which can be formatted as HFS+.
4. normalization
[available in ZEVO; must be specified at the time a dataset is created] Option to enable normalization=formD before putting any data in the pool. The same effect can be achieved in an old pool by creating a new dataset with normalization=formD, and copying data there. This is primarily useful to users who work in languages other than English.
5. case-sensitivity
[available in ZEVO; must be specified at the time a dataset is created] Option to enable casesensitivity=insensitive before putting any data in the pool. The same effect can be achieved in an old pool by creating a new dataset with casesensitivity=insensitive, and copying data there.

Pool properties
1. ashift
[available in ZEVO; must be specified at the time a vdev is created] Option to create the new pool with vdevs having ashift=12. If an old pool was created with the default of ashift=9, there is no way to change the old pool's vdevs to ashift=12, though new ashift=12 vdevs can be added alongside the preexisting ashift=9 vdevs using zpool add -o ashift=12.

Considerations other than properties
1. rebalancing
If the pool had new vdevs added after pool creation, data may be unevenly distributed amongst the vdevs. Creating a new pool with all vdevs in place before putting data on the pool should lead to a balanced distribution of data across the vdevs.
ilovezfs
 
Posts: 232
Joined: Thu Mar 06, 2014 7:58 am

Re: Futher life of Zevo volumes?

Postby dukzcry » Thu Jul 31, 2014 8:48 pm

ilovezfs,

Thank you for detailed info :-)
User avatar
dukzcry
 
Posts: 2
Joined: Sun Mar 23, 2014 2:38 am


Return to Non-OpenZFS Pools

Who is online

Users browsing this forum: No registered users and 4 guests

cron